That's an interesting thought. I agree that it'll cause a short-term disruption, but then soon enough people will find the best ways to optimize their earnings.
I'm fully in favour of making iterative changes, but iteration can only happen with... iterating. That requires making one change at a time, testing against a control, and studying the results of said change. Since HF16, we have had all kinds of things bundled in, and any empirical evidence to suggests what works becomes obfuscated.
Things are so broken here that I'm almost going to buy the argument that even a potentially bad change is good at this point. Almost...